(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Streptothricin F is a bactericidal antibiotic effective against highly drug-resistant gram-negative bacteria that interacts with the 30S subunit of the 70S ribosome [1]

['Christopher E. Morgan', 'Department Of Pharmacology', 'Case Western Reserve University School Of Medicine', 'Cleveland', 'Ohio', 'United States Of America', 'Yoon-Suk Kang', 'Department Of Pathology', 'Beth Israel Deaconess Medical Center', 'Boston']

Date: 2023-05

The streptothricin natural product mixture (also known as nourseothricin) was discovered in the early 1940s, generating intense initial interest because of excellent gram-negative activity. Here, we establish the activity spectrum of nourseothricin and its main components, streptothricin F (S-F, 1 lysine) and streptothricin D (S-D, 3 lysines), purified to homogeneity, against highly drug-resistant, carbapenem-resistant Enterobacterales (CRE) and Acinetobacter baumannii. For CRE, the MIC 50 and MIC 90 for S-F and S-D were 2 and 4 μM, and 0.25 and 0.5 μM, respectively. S-F and nourseothricin showed rapid, bactericidal activity. S-F and S-D both showed approximately 40-fold greater selectivity for prokaryotic than eukaryotic ribosomes in in vitro translation assays. In vivo, delayed renal toxicity occurred at >10-fold higher doses of S-F compared with S-D. Substantial treatment effect of S-F in the murine thigh model was observed against the otherwise pandrug-resistant, NDM-1-expressing Klebsiella pneumoniae Nevada strain with minimal or no toxicity. Cryo-EM characterization of S-F bound to the A. baumannii 70S ribosome defines extensive hydrogen bonding of the S-F steptolidine moiety, as a guanine mimetic, to the 16S rRNA C1054 nucleobase (Escherichia coli numbering) in helix 34, and the carbamoylated gulosamine moiety of S-F with A1196, explaining the high-level resistance conferred by corresponding mutations at the residues identified in single rrn operon E. coli. Structural analysis suggests that S-F probes the A-decoding site, which potentially may account for its miscoding activity. Based on unique and promising activity, we suggest that the streptothricin scaffold deserves further preclinical exploration as a potential therapeutic for drug-resistant, gram-negative pathogens.

Funding: This work was supported by R21 AI140212 to R.M. and J.E.K, and R01AI157208 to J.E.K., E.W.Y. and R.M. from National Institutes of Health. K.P.S. and K.E.Z. were supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under award numbers F32 AI124590 and T32 AI007061, respectively. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability: All relevant data are within the paper and its Supporting Information files with the following exception: Atomic coordinates and EM maps for SF bound to the A. baumannii P-site tRNA 70S, E-site tRNA 70S and Empty 70S have been deposited to the RCSB Protein Data Bank (PDB) under accession codes 7UVV, 7UVW and 7UVX and the Electron Microscopy Data Bank (EMDB) under accession codes EMD-26817, EMD-26818 and EMD-26819, respectively. Complexes of S-D with the A. baumannii P-site tRNA 70S, E-site tRNA 70S and Empty 70S have been deposited in the PDB under accession codes 7UVY, 7UVZ and 7UW1 and EMDB under codes EMD-26820, EMD-26821 and EMD-26822, respectively.

This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Based on potentially insufficiently explored therapeutic potential, we sought to further characterize the properties of S-F and S-D, the major constituents of nourseothricin, purified to homogeneity using modern techniques and authenticated by several complementary state-of-the-art methods. This analysis included in vitro and in vivo analysis and mechanism of action studies combining use of both mutational resistance and cryo-EM–based structural analysis. Taken together, our data establish S-F as a potently bactericidal natural product scaffold with a unique mechanism of action with what we believe are compelling properties for future medicinal chemistry exploration. To this end, we also separately established a diversity-enabling total synthesis for S-F, which is reported elsewhere [ 15 ].

Initial studies on toxicity in animals from the 1940s used only semiquantitative activity measures and impure compound with large doses presumably exceeding 100 mg/kg [ 8 ] and are therefore problematic to interpret. More recent studies in mice with fractionated streptothricins showed that toxicity varies with the length of the poly-β-lysine chain: specifically, a murine LD 50 of 300 mg/kg for S-F with 1 lysine compared to an LD 50 of approximately 10 mg/kg for S-D and S-C with 3 and 4 β-lysines, respectively [ 8 , 9 ]. For comparison, the murine LD 50 is 52 mg/kg IV for gentamicin [ 10 ]; 40 mg/kg intraperitoneally (IP) for colistin methanesulfonate [ 11 ]; and 260 mg/kg IP for tobramycin [ 12 ]. Streptothricin’s delayed toxicity towards renal proximal convoluted tubules, only demonstrated histologically in the literature for S-C in rats [ 13 ], has not been fully explained at a mechanistic level [ 8 ]. Of note, S-C at 40 mg/kg, well above the lethal dose, did not cause liver damage as assessed by serum aspartate aminotransferase (AST) and alanine aminotransferase (ALT) enzymatic assays [ 13 ], highlighting kidney damage as the primary limiting toxicity.

Streptothricins share streptolidine and carbamoylated gulosamine sugar moieties. They are distinguished by differing numbers of β-lysines attached end-to-end through amide bonds to the ε-amino groups. Nourseothricin is the natural product mixture of several streptothricins, predominantly S-F (1 β-lysine) and S-D (3 β-lysines). Acetylation of the β-amino group blocks activity and is the major known mechanism of antimicrobial resistance to streptothricins. S-D, streptothricin D; S-F, streptothricin-F.

Streptothricin is now known to be a natural product mixture, currently referred to as nourseothricin. For clarity, we will henceforth refer to the natural product mixture as nourseothricin and individual components by their specific streptothricin designation. Each streptothricin in nourseothricin consists of 3 linked components: (a) a streptolidine, a guanidine-lactam bicyclic 5–6 ring system; (b) a carbamoylated gulosamine; and (c) a homopolymer consisting of varying numbers of β-lysine residues, linked head to tail. The individual streptothricins are designated by letter suffixes (A-F and X), based on the number of lysine residues, with the most abundant streptothricins in the natural product, streptothricin F (S-F) and streptothricin D (S-D), having 1 and 3 lysine moieties, respectively (see Fig 1 ).

Here, we further considered the properties of a historic antibiotic scaffold. Streptothricin was originally isolated by Waksman and Woodruff in 1942 from a soil Actinomyces [ 3 ]. It generated intense initial excitement as the first identified antibiotic natural product with potent activity against gram-negative organisms and additional activity against Mycobacterium tuberculosis [ 4 ]. It was specifically noted to completely cure Brucella abortus infection in guinea pigs [ 5 ] and otherwise lethal Salmonella paratyphi B infection in mice [ 6 ], infections of concern at the time. Based on these attributes, it went into fermentative production at Merck. However, in a limited human trial (details not published) [ 7 ], the streptothricin natural product induced reversible kidney toxicity, and, therefore, was not further pursued as a therapeutic.

Moreover, there is little doubt that resistance will emerge to agents currently in the pipeline. We are therefore clearly in need of several new gram-negative agents that are unique in terms of antimicrobial class and potential vulnerabilities, and which can diversify our antimicrobial therapeutic portfolio [ 2 ].

The rapid emergence of antimicrobial resistance presents a significant challenge for treatment of bacterial infections. Carbapenem-resistant Enterobacterales (CRE) and Acinetobacter baumannii are of particular concern. Almost all approved antimicrobials that can overcome the gram-negative permeability barrier are natural products or synthetic or semisynthetic derivatives of natural products. Small molecules commonly available in high-throughput screening libraries rarely share similar physicochemical properties associated with gram-negative penetrance and activity [ 1 ]. Therefore, high-throughput screening efforts to identify novel antimicrobials using synthetic chemical libraries with rare exceptions have been nonproductive [ 1 ]. As a result, there is a significant antimicrobial discovery void.

While not seen in S-F, site 4 of S-D is located in the peptide exit tunnel ( S12B Fig ). The S-D molecule is stabilized via electrostatic interactions with A1612 of the 23S rRNA and residues K90, R92, A93, and R95 of rProtein L22. This site is in proximity of the binding site for macrolide antibiotics, which also occupy the peptide exit tunnel to interfere with peptide elongation. While these sites are not seen in the mutational experiments, sites 2 and 4 are located in vital locations of the ribosome, where the first is located in a position that may interfere with the transfer of deacylated tRNA to the E-site of the 50S and the second with the elongating peptide chain. These sites invite further experimentation. The remaining of the 9 other S-D binding sites (sites 1, 3, 5, 6, 7, 8, 9, 10, and 11) are shown in S12 Fig . As the 70S ribosome is a large complex, it is presumed to offer many binding sites that are potentially druggable. Our 70S–S-D structure indeed underscores the possibility that we can identify potential therapeutic strategies via this structural information. Of note, the orientation of A1492 and A1493, residues fixed in a flipped-out position during binding of 2-deoxystreptamine aminoglycosides and thought to stabilize binding of noncognate tRNAs [ 24 ], was not affected.

The second S-F-binding site is found to partially overlap with site 2 of S-D ( S11 and S12 Figs). This second S-F site is located at the proximity of the E-site in the 50S ribosomal subunit ( S11A, S11C and S12B2 Figs). The streptolidine and gulosamine moieties of S-F and S-D overlap at this site, interacting with H20 of rProtein L28, while the interactions of the β-lysine tails differ. Binding by these streptothricins is occluded by binding of tRNA at the 50S E-site ( S11B Fig ).

(A, B) Location of the S-F and S-D binding sites in the A. baumannii 70S ribosome with P-site tRNA, respectively. (C, D) Hydrogen-bonding interactions between the streptolidine moieties of S-F and S-D and C1054 of the 16S rRNA, respectively (E, F) Electrostatic interactions of the S-F and S-D 12-carbamoylated gulosamine moiety with A1196, respectively. (G, H) Electrostatic interactions between β-amino of the β-lysine and the O2’ atoms of C1054 and U1052. Throughout the figure, the 50S is blue, 30S is yellow, S-F is gray sticks, S-D is green sticks, and highlighted residues are cyan. The cryo-EM densities corresponding to the 2 terminal β-lysine moieties of S-D were very weak, indicating that these 2 lysine moieties are flexible. We therefore did not include these 2 terminal β-lysine moieties in panels D, F, and H. cryo-EM, cryo-electron microscopy; S-D, streptothricin D; S-F, streptothricin-F.

The structures of S-F and S-D bound to the 70S ribosome of A. baumannii were solved using single-particle cryo-electron microscopy (cryo-EM) ( S7 – S10 Figs, S7 and S8 Tables, S7 and S8 Data ). Binding of S-F and S-D were only observed in cryo-EM structures following incubation of ribosomes with 100 μM of each antibiotic, but not when ribosomes were soaked with a physiologically more relevant 10 μM concentration. Throughout the 70S, 2 distinct S-F-binding sites were identified for 70S–S-F, while a total of 12 S-D-binding sites were observed for the 70S–S-D complex ( S11 and S12 Figs). Notably, through comparison of both cryo-EM structures, only 1 S-F-binding site (site 1 of S11 Fig ) was found to completely overlap with one of the S-D-binding sites (site 12 of S12 Fig ). This site is located at the 30S head, specifically helix 34 of the 16S rRNA ( Fig 7A and 7B ). It is characterized by multiple hydrogen bonds between the streptolidine moiety of the streptothricin backbone and the 16S rRNA C-1054 nucleobase (E. coli rather than A. baumannii 16S rRNA base numbering is used; Fig 7C and 7D ) and by the gulosamine sugar and in particular the C-10-carbamoyl moiety with the A1196 base ( Fig 7E and 7F ). The β-amino group of the β-lysine also interacts electrostatically with O2’ of the riboses in U1052 and C1054 ( Fig 7G and 7H ). A further polar contact was made between the C-10-carbamoyl moiety and the carbonyl group of the rProtein s3 E161 amide bond. Surprisingly, the cryo-EM densities corresponding to the 2 terminal β-lysine moieties of S-D were very weak, indicating that these 2 lysine moieties are flexible. We therefore did not include these 2 terminal β-lysine moieties in the 70S–S-D cryo-EM structure. It was observed that both S-F and S-D utilize a similar mode of binding to interact with the 70S ribosome, but this binding mode is quite distinct from other antibiotics, such as eravacycline and negamycin, which are known to bind to this region ( S13 Fig ).

Despite overlap of identified nourseothricin resistance mutations with the reported binding pocket of tetracyclines in 16S rRNA helix 34 [ 23 ], biological data suggested that the mechanism of action of the bactericidal, miscoding-inducing streptothricin and bacteriostatic tetracyclines should be distinct. Supporting this notion, C1054A only conferred modest cross-resistance, and A1196C/G conferred negligible if any cross-resistance to tetracycline, doxycycline, minocycline, tigecycline, or eravacycline ( Table 1 ), respectively. In addition, expression of the ribosomal protection protein, TetM, had no effect on activity of either nourseothricin or the apramycin control (known to bind helix 44), while conferring expected high-level resistance to tetracycline and minocycline ( Table 2 ). Furthermore, a limited sampling of Staphylococcus aureus clinical strains expressing tetM gene, showed low nourseothricin MICs (0.5 μg/mL), while demonstrated the expected high MICs for tetracycline ( S6 Table ).

To identify potentially unique interactions with the ribosome, we sought to identify high-level nourseothricin resistance mutants by selecting for growth on media containing nourseothricin at 32 to 64 times the MIC. Two classes of high-level nourseothricin resistance mutants were obtained only in the single rrn operon, SQ110, E. coli mutant strain at a frequency of approximately 6.2 × 10 −9 but not in wild-type (wt) E. coli, which has 7 rrn operons, implying mutations were recessive. In total, 54, 2, and 5 independently isolated strains containing 16S rRNA C1054A, A1196C, and A1196G mutations, respectively, were identified. All grew noticeably slower than the SQ110 parent. In addition, 16S rRNA C1054T was identified in 16 strains that proved nonviable on passaging from frozen stocks and were therefore not further analyzed. In Table 1 , we show modal MIC values from 3 biological replicates for 2 randomly selected mutants of each type to rule out potential contributions of secondary mutations. C1054 and A1196 mutants conferred >256-fold and 16- to 32-fold increases in MIC for nourseothricin, respectively. Linkage of these mutations to the resistance phenotype and confirmation that these mutations are recessive were further supported by full restoration of susceptibility after introduction of a wt rrnB rRNA operon into the mutant strains, in contrast to absence of effect with the empty vector control ( S5 Table ). Furthermore, in vitro translation extracts prepared from the SQ110 C1054A mutant strain N1 were resistant to nourseothricin at the highest concentrations tested ( S6 Fig and S6 Data ). In contrast, extracts from parent SQ110 were susceptible with inhibition similar to observations with commercial E. coli K-12 extracts (Figs 3 and S6 ). Confirming that the resistance observed with the C1054A extract was nourseothricin specific, in vitro translation extracts from both parent SQ110 and strain N1 C1054A were similarly inhibited by the positive control, apramycin.

Therapeutic effect after a single dose of S-F was tested in the murine thigh model after infection with the pandrug-resistant Nevada CRE strain (AR-0636). For these studies, neutropenic mice were rendered mildly renal deficient with uranyl nitrate to more closely simulate human excretion kinetics [ 22 ]. Tissue was harvested 24 hours postinfection. In the absence of treatment, CFU increased >2log 10 . In contrast, S-F at 50 mg/kg or 100 mg/kg led to a greater than 5log 10 reduction in CFU with absence of detectable CFU in 3 of 5 mice treated with the higher dose ( Fig 6 , S5 Data ). Thus, substantial in vivo bactericidal treatment effect was observed with a single dose of S-F at levels with minimal or no observable toxicity.

To identify the single-dose maximum tolerated dose (MTD) of S-F and nourseothricin, CD-1 mice were injected IP with ascending doses of 0, 50, 100, 200, and 400 mg/kg S-F and 5, 10, 20, and 50 mg/kg nourseothricin, respectively (n = 4 per dose). Nourseothricin was used as a surrogate for S-D, based on the difficulty of isolation of the latter in quantities needed for the experiments. Over the next 72 hours, no signs of distress were observed with dosing up to 200 mg/kg of S-F and 20 mg/kg of nourseothricin, respectively. Two mice at 400 mg/kg of S-F and 1 mouse at 50 mg/kg nourseothricin died, or became moribund and were euthanized. Histology of kidneys was specifically examined for all mice based on described delayed renal toxicity in rats for the streptothricin natural product [ 13 ]. For S-F, proximal convoluted tubular damage was noted sporadically at 100 mg/kg and more diffusely at ≥200 mg/kg. For nourseothricin, proximal tubular damage was pronounced at ≥10 mg/kg. Glomeruli, distal tubules, and vasculature were spared ( Fig 5 ). Therefore, the maximum dose without observed pathological effect was 50 mg/kg for S-F and 5 mg/kg for nourseothricin, respectively.

J774 macrophages and LLC-PK-1 renal epithelial cells were treated with 2-fold doubling dilutions of nourseothricin, S-D, and S-F for up to 5 days in the presence of SYTOX-Green. SYTOX-Green is a cell membrane-impermeant nucleic acid binding dye that fluoresces on binding to nuclear DNA. It therefore provides a real-time readout of eukaryotic cell membrane permeabilization associated with cell death that can be continuously monitored through fluorescence measurements. Cytotoxicity was minimal to absent after a single day incubation but increased on subsequent days. Nourseothricin and S-D effects were essentially indistinguishable. S-F toxicity was only observed at molar concentration at least 10-fold greater than S-D beginning at 32 μM, significantly above MIC ranges observed in activity spectrum analysis. Each data point represents mean and standard deviation for assays performed in quadruplicate. Data are available in S4 Data . MIC, minimal inhibitory concentration; S-D, streptothricin D; S-F, streptothricin-F.

Eukaryotic cell cytotoxicity was determined against J774A.1 macrophage and LLC-PK1 proximal tubule kidney cells line during a 5-day incubation. Cytotoxicity was measured using a previously established SYTOX Green exclusion, real-time fluorescence assay [ 20 , 21 ]. Notably CC 50 for these cell lines was >10× higher for S-F than S-D ( Fig 4 , S4 Data ). Cytotoxicity was also relatively delayed for S-F compared with S-D, first apparent on day 2 of incubation, with more prominent cytotoxicity for both S-F and S-D noted after 5 days of incubation. The delay in observed cytotoxicity was more pronounced for LLC-PK-1 in which cytotoxicity for S-D and S-F were not appreciable until day 3, and of borderline magnitude for S-F at day 5 ( Fig 4 ). Notably, the lowest observable adverse effect level for S-F during prolonged 5-day incubation was 32 μM, substantially above the MIC 99 of 4 μM for CRE.

(A) Inhibition of prokaryotic in vitro translation using coupled in vitro transcription–translation extracts with readout from a nanoluciferase reporter. (B) Inhibition of eukaryotic in vitro translation using coupled in vitro transcription–translation extracts with readout from a nanoluciferase reporter. S-F, S-D, NTC, APR, and TET data represent mean and standard deviation from 3 independent experiments. Data are available in S3 Data . APR, apramycin; NTC, nourseothricin; S-D, streptothricin D; S-F, streptothricin-F; TET, tetracycline.

We measured in vitro action on ribosome translation (i.e., independent of bacterial permeability and efflux effects) using commercially (NEB) available prokaryotic (E. coli) and eukaryotic (rabbit reticulocyte) in vitro coupled transcription–translation systems with readout using nanoluciferase gene constructs ( S5 Fig ). In this analysis, we found that both S-F and S-D were approximately 40-fold more selective for prokaryotic than eukaryotic ribosomes based on IC 50 values determined from Hill slopes of dose–response curves ( Fig 3 , S3 Data). Interestingly, S-D was approximately 10-fold more potent than S-F on a molar basis, helping to explain the lower MICs noted for the former. Inhibition by apramycin and tetracycline was as expected. As the difference in translational inhibition by S-F and S-D was greater than the difference in their mean MICs, we, therefore, considered whether S-F might have relatively enhanced bacterial penetrance. However, the relative activity of S-D and S-F was not changed in isogenic E. coli lptD and tolC mutants ( S4 Table ), suggesting that the outer membrane permeability and efflux pump activity, as assessed by these measures, did not contribute to this discrepancy in activity observed in extracts and whole bacteria.

Bacterial killing by unfractionated, crude preparations of streptothricins was previously noted by Waksman and Woodruff for Escherichia coli and Micrococcus lysodeikticus after 17 and 65 hours of exposure without reference to an underlying MIC [ 3 ]. To characterize potential bactericidal activity, we therefore performed time-kill analysis for S-F and nourseothricin against the pandrug-resistant CRE K. pneumoniae Nevada isolate (AR-0636). Notably, we observed rapid bactericidal activity for both within 2 hours of exposure at 4X MIC without regrowth ( Fig 2 , S1 Data ). Findings were similar for carbapenem-resistant A. baumannii isolate MSRN 1450 ( S4 Fig , S2 Data). Therefore, bactericidal activity for highly resistant CRE and A. baumannii isolates were observed using current methods.

An LC–MS analysis of commercially purchased nourseothricin showed a composition of 65.5% S-F, 29.6% S-D, and 4.9% S-E (2 β-lysines). Therefore, the somewhat enhanced activity of S-D (mean MIC 0.44 μM) compared with nourseothricin (mean MIC 0.58 μM) for CRE likely reflects a combinatorial effect of the more potent S-D and less potent S-F, as well as contributions from other minor streptothricin congeners in the nourseothricin natural product mixture.

Using broth microdilution minimal inhibitory concentration (MIC) testing, we determined the MIC of S-F, S-D, and nourseothricin for a large group of multidrug-resistant CRE. These included strains expressing metallo- and/or serine carbapenemases and the pandrug-resistant, NDM-1–expressing Klebsiella pneumoniae Nevada strain, AR-0663 [ 18 ]. For these CRE (n = 39), MIC 50 and MIC 90 for S-F, S-D, and nourseothricin, respectively, were 2 and 4 (range 1 to 4) μM; 0.25 and 0.5 (range 0.25 to 2) μM; and 0.5 and 1 (range 0.25 to 2) μM. The average MIC was 5.6-fold greater for S-F than for S-D and 4.2-fold greater for S-F than for nourseothricin. We tested 4 A. baumannii strains in parallel with S-F, S-D, and nourseothricin (FDA-CDC #399–402) and found that the MIC for each was 4-fold greater for S-F than for S-D and nourseothricin. Data for individual isolates are listed in S3 Table . We also found that nourseothricin demonstrated broad activity against a diverse MDR/XDR A. baumannii strain set (n = 104), many isolates from wounded US soldiers [ 19 ]: MIC 50 = 2 μM; MIC 90 = 32 μM, MIC 99 = 32 μM; range <0.125 to 64 μM.

Our original studies with streptothricins were motivated by the observation that a nourseothricin resistance cassette served as a reliable genetic marker for experiments in otherwise highly drug-resistant gram-negative pathogens [ 16 , 17 ]. This led to the question of whether nourseothricin or its constituent streptothricin homologues had more general activity against multidrug-resistant gram-negative pathogens such as CRE and A. baumannii. We, therefore, adapted a previously reported method to purify S-F (1 β-lysine) and S-D (3 β-lysines) to homogeneity using Sephadex chromatography [ 9 ]. Confirmation of purity and precise molar content was determined by elemental analysis, NMR, and LC–MS as described in S1 Results ; S1 Materials and Methods ; S1 – S3 Figs; and S1 and S2 Tables.

Discussion

We demonstrate that streptothricins, in particular S-F, S-D, and the natural product mixture, nourseothricin, are highly active against contemporary, carbapenem-resistant E. coli, Klebsiella, Enterobacter, and A. baumannii. Their activity spectra are therefore highly relevant to the emerging gap in gram-negative antimicrobial coverage, even more so since they appear to act in a unique manor to kill bacterial cells.

Most of the existing description of streptothricin class antibiotic activity was from several decades ago and made use of unstandardized microbiological and physicochemical methods. In general, the natural product mixture was characterized as a whole without quantitative delineation of contributions from constituent streptothricins. Instead, here, we made use of modern preparative techniques and methods to characterize the activity of S-F and S-D in comparison with the natural product mixture currently available, gaining valuable insights into their respective properties.

We found that S-F maintained potent antimicrobial activity, although it was about 6-fold less potent than S-D. Importantly, S-F showed at least 10-fold lower toxicity than S-D and the natural product mixture, nourseothricin, in vitro and in vivo. Notably, in single, maximal tolerated dose experiments, treatment with the nourseothricin natural product mixture was associated with proximal tubule kidney damage at relatively low doses (10 mg/kg). In contrast, S-F showed a favorable therapeutic ratio with substantial therapeutic effect at 50 to 100 mg/kg to the point of single-dose sterilization of CRE infection with minimal or no renal histopathological toxicity.

Biologically, streptothricins demonstrate activity similar to 2-deoxystreptamine (2-DOS) containing aminoglycosides such as gentamicin, amikacin, tobramycin, and apramycin, both inhibiting protein synthesis with rapid, potent bactericidal activity [25], as observed here for CRE and A. baumannii, and inducing translational miscoding [14,26]. It was therefore an original expectation that streptothricins would interact with the 30S ribosomal subunit target of aminoglycosides in a similar manner. However, the combination of structural and resistance studies supported an alternative and unique mechanism of action.

The 2-DOS aminoglycosides are known to bind helix 44 in 16S rRNA (residues 1400 to 1410 and 1490 to 1500), forming a binding pocket near the ribosome decoding center [27–31] (S14C Fig). In doing so, they induce a flipping out of 16S rRNA bases, A1492 and A1493. This molecular rearrangement is either associated with or stabilizes binding of noncognate tRNA to the A-site leading to miscoding [32]. Consistent with their binding profile, the activity of 2-DOS aminoglycosides is blocked by methylation of residues G1405 and A1408, as well as by recessive, resistance-conferring point mutations at positions 1406, 1408, 1409, 1491, and 1495 [33–39]. However, despite demonstrating miscoding and rapid bactericidal activity, characteristics of aminoglycosides, neither methylation of G1405 and A1408 by ArmA and NpmA resistance enzymes (S9 Table), nor the A1408G and G1491A mutants (S10 Table) tested, affected nourseothricin activity (S1 Results).

Instead, our structural and resistance studies identified an interaction with helix 34 distinct from other known translation inhibitors. Specifically, S-F makes direct polar contacts with nucleobases, C-1054 and A-1196. This contrasts with the stacking interactions of tetracyclines [40] with C-1054 (S13 Fig) and interaction of the pseudodeptide antibiotic, negamycin, solely with phosphoribose backbone of helix 34 (S13 Fig) [41,42]. Of interest, bonding of streptolidine with C1054 (S14A Fig) resembles G:C Watson–Crick base pairing (S14B Fig). This observation likely accounts for high-level resistance conferred by nucleobase changes, which would interrupt this strong bonding interaction, and therefore cannot be tolerated. Similarly, 2 hydrogen bonds between the carbamylated gulosamine and the A1196 nucleobase likely account for the midrange resistance levels associated with mutation at this position. In contrast to the strong effects on streptothricin activity, C1054A and A1196C/G mutations had negligible effect on tetracycline, doxycycline, minocycline, tigecycline, and eravacycline MIC values. In addition, the streptothricin resistance mutations were distinct from previously isolated mutations conferring strong resistance to tetracyclines, at 16S rRNA U1060A and G1058C [43–45]; spectinomycin at A1191 and C1192 [31,33,39,46,47]; and negamycin, at U1060A, U1052G, A1197U [41,42], consistent with known contacts in structural studies, and further supporting a distinct biological mechanism of action. Lastly, the ribosomal protection protein, TetM, which blocks access to the tetracycline binding site [48], also had no effect (Table 2). Structural studies indicated that P591 in loop 3 of domain IV of TetM makes stacking interactions with C1054, similar to tetracyclines. However, superimposition of the S-F and Tet-M ribosome structures suggest lack of steric clash of TetM with the S-F/S-D binding site, providing a structural rationale for TetM’s lack of effect on streptothricin activity (see S15 Fig).

Our structural data further explain streptothricin acetyl transferase-based resistance to S-F and S-D associated with acetylation of the β-amino group of the first β-lysine attached to the gulosamine sugar. Specifically, our modeling indicates that acetylation cannot be accommodated in the S-F binding pocket without steric clash with the phosphoribose backbone of U1052 and C1054 (Fig 7F). Finally, our data help explain several observations related to activity of naturally occurring S-F analogs. In S-F, the hydrogens on C-2 and C-3 of the streptolidine β-lactam ring are located in trans (2S, 3S configuration). Although comparative analysis between naturally occurring S-F analogs has not been systematically performed, an S-D analog with a cis-fused streptolidine lactam moiety (2R, 3S) showed a 32-fold increase in MIC for B. subtilis [49] compared to a previous report for S-F by the same group [50]. The disruption of the stereochemistry at the interface of the critical binding interactions of the streptolidine and C-1054 presumably accounts for this loss of activity, analogous to the disruption of this interaction in the C1054A mutant. In contrast, methylation of the nitrogen in the streptolidine lactam ring, with, in the analog, albothricin [51], or without, in the analogs, A37812 [52] and N-methyl S-D [53], removal of the C-4 hydroxyl group from the streptolidine ring, both positions outside of the interface with C1054, did not appear to undermine activity. Furthermore, carbamoylation at the C-12 (see Fig 1) rather than C-10 position of the gulosamine ring (analog A-269A’) was associated with an approximately 16× increase in MIC [50], presumably because of disruption of contacts with A1196, analogous to the disruption of the same interactions in the A1196C/G mutants (Table 1). The amino terminus (ε-amino group) of the β-lysine appears unbonded and surface exposed, allowing for the presumably relatively unrestricted addition of additional β-lysines, e.g., in S-D. Interestingly, substitution of the β-lysine with the α-alpha amino acids, methyl-glycine (named alternatively A-269A [54], SF-701 [55], or LLBL-136 [56]), or methyl-imino glycine (named BD-12 [57] or LL-AB664 [58]), were tolerated and may even have superior activity. Modeling suggests new contact of the α-amino group of these amino acids with the C1054 ribose, replacing the contacts of the β-lysine β-amino group with the C1054 phosphate and U1052 ribose. Intriguingly, it is possible that existing streptothricin acetyltransferases may be inactive against these α-amino acids substitutions, an area of investigation not yet reported in the literature, and, therefore, these naturally occurring modifications might extend activity spectrum to otherwise resistant organisms.

Regarding potential mechanisms of action, previous structural studies have described stacking interactions between C1054 and the wobble nucleotide of the A-site tRNA, anti-codon stem loop [59]. Notably, superposition of the S-F structure onto C1054 in the PDB structures of E. coli (PDB 7K00 [60]) and T. thermophilus (PDB 4V8B [61], 3T1H [62], 1IBM [24], 4WQ1 [63], 6XHV [61], and 4V8B [64]) ribosome structures including various A-site tRNA and mRNA codons suggests steric interaction with the wobble nucleotide of the anti-stem loop structure (S14C Fig) and potential probing of the minor groove between the codon and anticodon. Intriguingly, too, C1054A was the first known ribosomal suppressor mutation of the UGA stop codon [65–68] and is associated with readthrough of all 3 stop codons [66], supporting a critical role of this nucleobase in coding fidelity. These observations lead to a hypothetical mechanism explaining both observed miscoding and translation inhibition by S-F through effects on tRNA binding to the A-decoding site. Specifically, S-F may directly interact with the third, wobble position, and stabilize noncognate A-site tRNA interactions, a mechanism of action previously proposed for negamycin [69]. It should be noted that we have thus far not solved structures with streptothricins bound to translationally active ribosome complexes including both A-site and P-site tRNAs and mRNA. These complexes would provide further insights into mechanism of action.

It was theorized previously that the observed concentration dependence of S-F induced miscoding might be a consequence of multiple binding sites with presumably synergistic or additive mechanisms of action [14]. However, the other binding sites of S-F and S-D identified in cryo-EM structures in the 70S ribosome appear not to confer significant antimicrobial activity on its own, based on minimal, if any, retained activity when the 30S binding site is disrupted in the C1054A mutant. Nevertheless, these results do not rule out contribution to antibacterial activity once the primary 30S binding site is engaged. Notably, the supraphysiological 100 μM concentration of streptothricin used for cryo-EM studies (streptothricins were not resolved when soaking at 10 μM concentrations) likely contributed to the numerous additional and presumably nonspecific and physiologically irrelevant secondary binding sites observed.

Of interest, the nucleotides corresponding to 16S rRNA C1054 and A1196 in human 18S rRNA (PDB 7TQL; [70]) are conserved and occupy similar flipped out, stacked positions in helix 34 (S16 Fig). Likewise, the bacterial G530 from helix 18, which is also predicted to make contact with the streptolidine ring on the side opposite of its interaction with helix 34, and its human counterpart occupy nearly identical positions. Although bacterial s3.E161 is replaced by s3.Q145 in the human s3 rProtein, the position of their respective amide bond C-terminal carbonyl groups, which is predicted to make polar contact with the streptothricin C-10-carbamoyl moiety, is conserved. Therefore, it is not clear what accounts for the relative selective inhibition of prokaryotic translation. It is possible that the sum of small deviations in these respective structures and others not appreciated contribute to this selectivity. Solving structures of streptothricins bound to the human ribosome may help address some of these possibilities.

In summary, we present data for compelling bactericidal activity of S-F against contemporary multidrug-resistant CRE and A. baumannii pathogens with in vivo confirmation of efficacy against an emblem of gram-negative antibiotic resistance. We also find evidence for unique mechanism of action targeting 16S rRNA helix-34 distinct from other known translation inhibitors. We therefore believe that further early-stage exploration of the historic scaffold is warranted with the ultimate goal of identification of analogs with potential for therapeutic development.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002091

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.